Learning Bayesian Statistics cover image

#48 Mixed Effects Models & Beautiful Plots, with TJ Mahr

Learning Bayesian Statistics

00:00

Is There a Problem With Your Prior Predictive Samples?

When you're losing regression, you need a link function. That's what the log ods are also callede logistic function. And of course, this distorts the parimeter space, which is on the real line,. It distorts it into e the zero one space, so that it beganme it becomes a probability. So if you're using classic normal, zero, ten prior, for instance, you get very, very implausible prior predictive samples. The sampler then complains about divergences or other issues. And so then you tune the priors, and you look at the prior predictive stuff.

Transcript
Play full episode

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app